Pulsed lasers are used extensively for material processing applications such as machining, drilling, and marking. In many of these applications identical beam propagation properties and the same laser pulse energy from pulse to pulse are a requirement. A pulse repetition rate and a pulse duration that are optimum for an operation on any one material will usually not be optimum for another operation or another material. Accordingly, an “ideal” pulsed laser would have independently variable pulse-repetition frequency (PRF) and pulse-duration to allow an optimum combination to be selected for most applications on most materials. Such an ideal laser could be termed as a “pulse-on-demand” laser for delivering a single pulse or a burst of pulses with an arbitrarily interval between pulses. This interval can range from a few microseconds (μs) to a second or greater. Consistent beam propagation properties are required so that the laser pulses can be consistently focused into a desired spot size at a predetermined location. Preferably PRF should be variable without varying the pulse duration.
One type of laser apparatus in which the PRF can be varied without a variation in pulse duration is a fiber-MOPA. In such apparatus seed pulses are generated by a modulated single-mode semiconductor diode laser or a continuous-wave (CW) laser followed by a modulator such as an electro-optic (E-O) modulator or an acousto-optic (A-O) modulator. Such a fiber-MOPA can be operated at a PRF from less than 100 kilohertz (kHz) up to about 5 megahertz (MHz) or greater, with a pulse duration selected between about 0.01 nanosecond (ns) and 100 ns or greater. The seed pulses are amplified by a chain of fiber-amplifier stages. Early stages are usually designated as pre-amplifier stages and subsequent stages are designated as power amplifier stages. Fiber-MOPAs for the above mentioned applications usually emit pulses having a wavelength between about 1000 and 1100 nm. The wavelength of the pulses can be shortened by harmonic-conversion or sum-frequency mixing in one or more optically nonlinear crystals.
Typically most of the amplifiers are energized by CW pump radiation. Because of this, there are certain factors which affect pulse stability at the output of a fiber-MOPA and the output characteristics of the fiber-MOPA can change significantly when a pulse operating regime is changed.
One factor relates to transient-gain oscillations in the fiber amplifiers when average power of the signal or a pulse energy changes. In this case, the fiber-MOPA provides variation in gain, typically in the first few milliseconds after switching from one regime to another. In Yb-doped fiber amplifiers this can be within about 1 to 3 milliseconds (ms) after switching, depending on the number of fiber amplifier stages in the amplifier chain. The more fiber amplifiers in the chain the longer is the transient gain oscillation time.
Typically, those amplifiers operate in a saturated regime when incoming optical pulses change the inversion population and gain of the amplifier. Most of the pre-amplifier stages operate in a mode wherein gain is saturated by the average power of a signal train. By way of example, saturation power in Yb-doped fiber with 6-micrometer (μm) core-diameter is between about 5 and 20 milliwatts (mW), depending on a signal wavelength. Typical average power from a diode-laser seed source with 10-ns pulses at a PRF of 200 kHz is about 2 mW. If there is a sufficiently long time interval between pulses or bursts of pulses (longer than 1% of the excited-state lifetime, which is about 700 microseconds (μs) in a typical Yb-doped silica fiber), then inversion population starts to change between pulses due to competition between amplified spontaneous emission and the CW pumping, and may cause a variation in the gain (and accordingly power) from pulse-to-pulse.
FIG. 1 is a reproduction of an oscilloscope trace schematically illustrating the form of the envelope of an original stable pulse-train having 5-ns pulses at a PRF of 200 kHz and the envelope of a new pulse-train having 1-ns pulses at a PRF of 1 MHz, around the time of switching from the original train to the new train, in a prior-art fiber MOPA. The time base is 200 microseconds-per-division on which scale individual pulses are not discernable. The severe fluctuation of peak power after switching from one train to the next is clearly evident. It can be seen that about 800 μs is required before the power begins to stabilize, during which time 800 individual pulses have been delivered each having a different peak power and pulse-energy.
In fiber-MOPA apparatus in which harmonic-conversion or sum-frequency mixing (frequency-conversion) is employed to provide shorter wavelength pulses, such changes in power can affect the efficiency of the frequency conversion and the accuracy of beam pointing. The efficiency of frequency conversion in an optically nonlinear crystal depends on phase-synchronism of interacting beams, and depends on the temperature of a nonlinear crystal. Variations of the crystal temperature detune the phase-synchronism from an optimal position and reduce the efficiency of the frequency-conversion process. Accordingly it is important to maintain the nonlinear crystal at a constant temperature. The presence of a small amount of absorption in the crystal and absorption of scattered signal light in the crystal holder leads to changes of the crystal temperature depending on incident average power. Because of this, the above described fluctuations in power from pulse to pulse in the fundamental fiber-MOPA can cause corresponding fluctuations in conversion efficiency and beam pointing in frequency conversion stages. Accordingly, there is need for a mode-of operating a fiber-MOPA that can mitigate, if not altogether eliminate, output pulse instability when a pulse regime is changed.